Carefully conducted and appropriately reported clinical trials are of fundamental importance in evidence-based medicine. This applies to both new and established treatments (see, for example, p291–2 of this issue). It is estimated that about half of all clinical trials are not reported; this is a problem of which we have been aware since 1986, if not earlier. Perhaps unsurprisingly, there is a bias to the publication of ‘positive’ trials. The selective reporting of trials to regulatory agencies and in the medical literature influences decisions about the licensing and use of pharmaceutical agents in clinical practice.